manage/uuspider/analysis/detailAnalysis_kind.py

# coding=utf-8
'''
Created on 2016年2月26日
对类目进行分析
@author: ChenHao
'''

class DetailAnalysisKind(object):
    
    def __init__(self, out_list_input):
        if out_list_input is not None:
            self.out_list = out_list_input
        else:
            self.out_list = []
    
    def _is_in_out_list(self, kindName):
#         print self.out_list
        if len(self.out_list) == 0:
            print("none")
            return False
        else:
            for obj_kind in self.out_list:
                if obj_kind["nameCn"] == kindName:
                    return True
            
        return False
    
    def _get_id_form_out_list(self, kindName):
        for obj_kind in self.out_list:
            if obj_kind["nameCn"] == kindName:
                return obj_kind["id"]
        return False
    
    def _add_one_into_out_list(self, kindName, parentid, level):
        id_kind = len(self.out_list) + 1
        d = dict()
        d["id"] = id_kind
        d["nameCn"] = kindName
        d["parentid"] = parentid
        d["level"] = level
        
        self.out_list.append(d)
    
    def _add_into_out_list(self, kind_list):
#         print "type:", type(kind_list)
#         print kind_list
#         print kind_list[0]
        for i, kindName in enumerate(kind_list):
            # 先检查自己在不在数据集合里面
#             print kindName
            if not self._is_in_out_list(kindName):
                # 检查自己的层级
                if i == 0:
                    # 如果是顶级
                    self._add_one_into_out_list(kindName, 0, 1)
                else:
                    # 先找父级
                    parentName = kind_list[i - 1]
                    parentid = self._get_id_form_out_list(parentName)
                    self._add_one_into_out_list(kindName, parentid, i + 1)
    
    def _get_mouser_kindid_by_kindName(self, kindName, kinds_list):
        for obj_kind in kinds_list:
            if kindName == obj_kind["nameCn"]:
                return obj_kind["id"]
            
    def _get_uu_kindid_by_kindName(self, kindName, kindsmouser, translateRelation):
        for obj_kind in kindsmouser:
            if kindName == obj_kind["nameCn"]:
                mouserid = obj_kind["id"]
                for o in translateRelation:
                    if mouserid == o["mouserid"]:
                        return o["uuid"]
    
    # 获得父级类目
    def _get_parent_by_parentid(self, parentid):
        for kind in self.out_list:
            if kind["id"] == parentid:
                return kind
        
        return None
        
    # 根据叶子节点id还原整条路径    
    def _get_parents_by_leafid(self, leafid):
        kind_list = list()
        kind_leaf = None
        for kind in self.out_list:
            if kind["id"] == leafid:
                kind_leaf = kind
        
        kind_list.append(kind_leaf)
        kind_parent = self._get_parent_by_parentid(kind_leaf["parentid"])
        
        while(kind_parent):
            kind_list.append(kind_parent)
            kind_parent = self._get_parent_by_parentid(kind_parent["parentid"])
        
        if len(kind_list) == 0:
            return None
        else:
            return kind_list
            
    # 最后为结果设置isLeaf字段，并设置
    def _set_isLeaf(self):
        #1、得到所有的parentid
        parentid_set = set()
        for kind in self.out_list:
            parentid_set.add(kind["parentid"])
        
        #2、遍历结果集并设置状态
        for kind in self.out_list:
            if kind["id"] in parentid_set:
                kind["isLeaf"] = False
            else:
                kind["isLeaf"] = True
                # 组装
                kind_list = self._get_parents_by_leafid(kind["id"])
                kind_list.reverse()
                uuidForCmp = ""
                for kd in kind_list:
                    uuidForCmp += kd["uuidForKind"]
                # 补全不足的位数
                uuidForCmp += "0" * (9 - len(uuidForCmp))
                kind["uuidForCmp"] = uuidForCmp
                
                
                
                
    # 最后为结果设置detno, uuidForCmp字段
    def _set_detno_and_uuidForCmp(self):
        #1、得到所有的parentid
        parentid_set = set()
        for kind in self.out_list:
            parentid_set.add(kind["parentid"])
        
        #2、遍历结果集并设置状态
        for parentid in parentid_set:
            same_parentid_count = 0
            for kind in self.out_list:
                if kind["parentid"] == parentid:
                    kind["detno"] = same_parentid_count + 1
                    if kind["level"] == 1:
                        kind["uuidForKind"] = str(same_parentid_count + 11)
                    elif kind["level"] == 2:
                        kind["uuidForKind"] = (str(same_parentid_count + 1)).zfill(3)
                    else:
                        kind["uuidForKind"] = (str(same_parentid_count + 1)).zfill(4)
                     
                    same_parentid_count += 1
                    
            
            
    
    def craw(self, db):
        detail_list = db.detail_json.find()
        for detail in detail_list:
            self._add_into_out_list(detail['kinds'])
        return self.out_list
            
if __name__ == '__main__':
    obj_spider = DetailAnalysisKind()
    out_list = obj_spider.craw()